The Agent Skills Directory

[PROMPT_INJECTION]: The skill describes tools for news sentiment analysis and news-based decision making (tools/news_sentiment.py). This ingestion of untrusted external data sources creates a surface for indirect prompt injection, where instructions hidden in news content could attempt to influence the agent's trading logic or research outcomes.
Ingestion points: External news data and sentiment analysis tools (news_sentiment.py, news_analysis).
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions for external data content.
Capability inventory: The system includes capabilities for autonomous trading (auto_trader.py), order execution (trade_executor.py), and strategy backtesting.
Sanitization: No explicit sanitization or validation of the external news text is described in the provided skill context.
[COMMAND_EXECUTION]: The architecture utilizes subprocess spawning from a Rust backend to run Python scripts across different virtual environments. The backtest_strategy tool accepts a strategy_code string, which suggests dynamic execution of code provided to the tool. While this aligns with the primary purpose of a quantitative research environment, it represents a capability for executing arbitrary code logic.

ai-quant-engineering